Configuration

The InfraPrism SDK can be configured through environment variables, client initialization parameters, or a combination of both.

Environment Variables

Variable	Description	Required
INFRAPRISM_API_KEY	Your InfraPrism API key	Yes
OPENAI_API_KEY	OpenAI API key (for OpenAI provider)	Provider-specific
ANTHROPIC_API_KEY	Anthropic API key (for Anthropic provider)	Provider-specific
INFRAPRISM_ENVIRONMENT	Environment tag (e.g., production, staging)	No
INFRAPRISM_DISABLED	Set to true to disable tracking	No
INFRAPRISM_DEBUG	Set to true for debug logging	No

Client Initialization

OpenAI

from infraprism import InfraPrismOpenAI

client = InfraPrismOpenAI(
    # OpenAI configuration
    api_key="sk-...",                    # OpenAI API key
    organization="org-...",              # OpenAI organization (optional)
    base_url="https://api.openai.com/v1", # Custom base URL (optional)
    timeout=30.0,                         # Request timeout in seconds
    max_retries=2,                        # Number of retries

    # InfraPrism configuration
    infraprism_api_key="ip-...",         # InfraPrism API key
    infraprism_environment="production",  # Environment tag
    infraprism_disabled=False,            # Disable tracking
    infraprism_debug=False,               # Enable debug logging
)

Anthropic

from infraprism import InfraPrismAnthropic

client = InfraPrismAnthropic(
    # Anthropic configuration
    api_key="sk-ant-...",                # Anthropic API key
    timeout=30.0,
    max_retries=2,

    # InfraPrism configuration
    infraprism_api_key="ip-...",
    infraprism_environment="production",
)

Configuration Options

infraprism_api_key

Your InfraPrism API key. Get one from the dashboard.

client = InfraPrismOpenAI(infraprism_api_key="ip-your-key")

Or use an environment variable (recommended):

export INFRAPRISM_API_KEY="ip-your-key"

infraprism_environment

Tag all requests with an environment name. Useful for separating development, staging, and production costs.

client = InfraPrismOpenAI(infraprism_environment="production")

The environment is automatically added as a tag and can be filtered in the dashboard.

infraprism_disabled

Completely disable InfraPrism tracking. Useful for local development or testing.

client = InfraPrismOpenAI(infraprism_disabled=True)

Or via environment variable:

export INFRAPRISM_DISABLED=true

When disabled, the SDK behaves exactly like the native OpenAI/Anthropic client with zero overhead.

infraprism_debug

Enable debug logging to troubleshoot integration issues.

client = InfraPrismOpenAI(infraprism_debug=True)

Debug mode logs:

• Event payloads being sent
• Batch upload timing
• Any errors in the tracking pipeline

Async Configuration

The async clients accept the same configuration options:

from infraprism import AsyncInfraPrismOpenAI

client = AsyncInfraPrismOpenAI(
    infraprism_api_key="ip-...",
    infraprism_environment="production",
)

# Use with async/await
response = await client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Hello"}],
    entity_type="customer",
    entity_id="acme-corp",
)

Request-Level Configuration

You can override configuration on a per-request basis:

# Override entity and tags for this specific request
response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Hello"}],
    entity_type="project",           # Override entity type
    entity_id="internal-tools",      # Override entity ID
    tags={"priority": "low"},        # Request-specific tags
    infraprism_disabled=True,        # Skip tracking for this call
)

Configuration Precedence

Configuration is resolved in this order (highest to lowest priority):

1. Request-level parameters
2. Client initialization parameters
3. Environment variables
4. Default values

Next Steps

• OpenAI Provider - OpenAI-specific configuration
• Anthropic Provider - Anthropic-specific configuration
• Azure OpenAI - Azure configuration